*INTLJEVIDENCEPROOF 100 Abstract. This article examines the importance of establishing a coherent system of safeguards against the risks of mistaken eyewitness identification. Memories of previously observed events are susceptible to distortion during encounters between a witness and other persons. It is argued that pre-trial procedures conducted under controlled conditions provide the most effective means of averting the risk of error and subsequent miscarriage of justice. However, these are of little value without a system of safeguards that (1) ensures that the procedures are conducted whenever the risk of error arises, (2) provides adequate regulation of what might occur before such procedures are conducted, and (3) ensures adherence to its prescriptions. These considerations provide the basis for a critical evaluation of some aspects of relevant procedure in England and Wales.
In a thought-provoking essay, ‘Identification and Misidentification in Legal Processes: Redefining the Problem’,1 William Twining challenged the orthodox view, that the problem of misidentification is something to do with the unreliability of eyewitness testimony in jury trials, as presenting an artificially narrow definition of the problem. Future research and public debate about the problem of identification in legal processes, he claimed ‘would benefit from being set in the context of a comprehensive model of legal processes, and of a clearly articulated, integrated theoretical framework’.2 This would entail divergence from the expository traditionof legal scholarship, the primary concern of which are the rules of evidence,3 and would transcend disciplinary boundaries. The expository approach is introspective, characterised in its most polarised form, by an analysis of problems of evidence that takes the evidential rules as the source *INTLJEVIDENCEPROOF 101 of both the problem and the solution with scant regard to the body of empirical facts lying beyond the corpus of those rules. Twining suggested that an analysis of identification ought to be carried out at a level of abstraction that ‘combines elements of contextual perspectives on legal process and some notions about information processing borrowed from cognitive psychology’. The criminal procedure relating to identification evidence still awaits this kind of attention and analysis. Twining observed that many of those who endeavour to resolve the problem of mistaken identification appear to adopt the view that the solution lies in effecting improvements in the rules concerning the manner in which identification parades are conducted.
Carson and Bull4 have suggested that the potential contribution of psychology to the discipline of law lies in encouraging the lawyers to rethink issues and challenge prevailing rules and practice. The relevant processes, both legal and cognitive, are complex and warrant greater attention than can be devoted to them here. This article is a rather ambitious attempt to demonstrate that psychologists' discoveries concerning the functioning of memory suggest the need for a cohesive system of legal safeguards. The principal objective is to demonstrate that a broad contextual evaluation of English procedure dispels any notion that the safeguards it currently affords offer robust protection against the risk of miscarriage of justice. The premise of the arguments developed below lies in the extent to which a witness's memory is susceptible to distortion as a result of encounters with other persons following the witness's observation of events giving rise to criminal proceedings.
The range of factors that might affect the accuracy of any identification attempted by an eyewitness is substantial. Numerous commentators have formulated taxonomies5 which include characteristics inherent in the witness, such as age,6 respective ethnicity of witness and suspect,7 and degree of physiological arousal *INTLJEVIDENCEPROOF 102 (as might occur when a violent event is witnessed).8 Environmental factors also have the potential to affect accuracy, for example, lighting conditions when the culprit is observed, the length of time that the witness has the culprit in view,9 whether the offence involved the use of a weapon.10 In providing structure to what would otherwise be a mere catalogue of factors, Wells'11 categorisation of factors as either system variables or estimator variables is of considerable importance for the evaluation and development of criminal procedure.
Estimator variables describe those factors capable of affecting identification accuracy that have an existence independent of anything that might be done by those responsible for criminal justice system policy, and of the practices pursued by those working in the system. Environmental circumstances, and the physiological characteristics of witness and suspect, fall into this category.
In contrast, the term ‘system variable’ describes a phenomenon believed to affect identification accuracy adversely which arises as a consequence of some conduct engaged in by agents (essentially police officers) in the pre-trial process. These include various forms of suggestive conduct engaged in by the agent while following the prescriptions of formal rules or exercising discretion conferred for the purposes of investigation.
In the context of estimator variables, so named because they are not controllable and their influence can only be ‘estimated’,12 mitigating the risk of miscarriages of justice requires an awareness of their potentially adverse effect on identification *INTLJEVIDENCEPROOF 103 accuracy. To employ a medical analogy; if we perceive mistaken identification as a disease, which in the criminal justice system we want to eradicate, there are two strains. We are powerless to prevent the ‘estimator strain’, the sources of which are the physical circumstances in which a culprit was observed committing an offence or the physiological characteristics of the actors involved. The most that the criminal justice system can do is to remain vigilant as to its various symptoms and implement an effective screening programme in an attempt to detect possible outbreaks of the disease.
The ‘system strain’ of the disease is a rather different matter. While the disease remains difficult to diagnose outbreaks are caused by the practices and procedures followed in the criminal process. It is, therefore, possible to take preventative measures by adopting and adhering to appropriate regimens concerning the treatment of eyewitness identification. Wherever there is interaction between a number of witnesses, or police officers and witnesses, there is a danger that a witness's memory and recollection of relevant events will be distorted.
There is a consensus among psychologists that the mind does not operate like a video recording, whereby once events are stored in the memory they can be recollected by accessing the discrete ‘part’ of the memory in which they are stored. Memory processes appear much more complex than this. Researchers have proposed models which differentiate different kinds of memory involving the processing of information between the various modes.13 One manifestation of malfunctioning in this process is the apparent malleability of memory.
One study, in which participants observed a stranger whom they were later asked to describe, provides a vivid illustration of the danger of semantic suggestion on witness recall.14 In the study, one group of witnesses were told by the interviewer that the man was a truck driver while another group were told that he was a dancer. Those who were told he was the former estimated his weight to be significantly greater than those who were told he was the latter. Similarly, where interviewers described the stranger as a ‘man’ estimates of his age were significantly higher than when he was referred to as a ‘young man’. As Kohnken has explained, the interviewee who experiences an event is unlikely to be able to perceive and encode all available information.15 Any additional information *INTLJEVIDENCEPROOF 104 acquired from an interviewer might lead the witness to use general knowledge and fall back on social expectations to fill ‘gaps’ in his recollection of events.
Perhaps of greater concern is that, where a witness is provided with incorrect information regarding a previously observed event or culprit, it appears that the manner in which information is processed can result in the witness recalling the erroneous information as a fact he observed during the original event.16 In a study conducted by Crombag and others,17 a number of people were interviewed about an incident in which a cargo plane crashed into a block of apartments in Amsterdam. Sixty per cent of the subjects stated that they had seen the plane colliding with the building on television and provided answers to detailed questions about the broadcast material, despite the fact that no one had captured any footage of the crash occurring. The subjects appeared to have pieced together what they heard about the crash from various sources, constructed a composite mental image of events surrounding the disaster and were subsequently confident (and probably convincing) in their claims to have watched it on television.
Questions remain as to whether, once a witness's memory has been distorted by the provision of ‘misinformation’ following observation of an event, the unadulterated memory of the original event can be retrieved. The view adopted in some quarters is that some memories undergo irreversible transformations.18 This cognitive frailty might pose a danger of mistaken identification occurring at numerous points in legal proceedings. Any interaction between witness and police officer, or a witness with other witnesses to the same event, provides an opportunity for an exchange of information or a situation in which a witness might feel obliged to make an identification or be encouraged to do so with less caution than might otherwise be the case. This proposition is supported by studies which suggest that misleading information given to a witness influences decisions made during identification procedures and gives rise to an increased risk of misidentification.19
*INTLJEVIDENCEPROOF 105 Further dangers lie in the social pressure and suggestion that flow from a witness's perception of the attitudes, conduct and expectations of other witnesses and the police,20 or the provision of information not perceived by the witness during the original event. The fact that the police arrest a suspect and request a witness to attempt an identification is inherently suggestive. It is reasonable, it is claimed, to expect that a witness will be inclined to act on the inference that the police believe they have the culprit. In such circumstances the witness will probably be less cautious and more willing to make a positive identification. Consequently it is likely that identifications made under such conditions are more susceptible to error.21 The problem might be exacerbated if the witness were to be pressed by police officers to take part in identification procedures following the arrest of a suspect. However, the pressure on a witness to make an identification need not be confined to circumstances in which a suspect has been arrested. Where police officers appear to the witness to be expending considerable effort or exuding particular enthusiasm in their efforts to track down a culprit, the witness may feel considerable social pressure to ‘do his bit’ and make an identification when the police arrest a suspect.
If the risk of miscarriages of justice occurring because of mistaken identification is to be reduced, then it is necessary that those who are responsible for policy and developing prescriptive procedures, and those who have responsibility for evaluating the sufficiency of identification evidence for various purposes, are aware of the existence of the two sets of variables (system and estimator), the potentially adverse effect that they might have on accuracy, and are sensitive to the implications for policy and practice.
The criminal process
In most cases22 the trial is the culmination of a legal process comprising a series of decisions and interactions engaged in by various parties to the proceedings. As Twining notes:
… by the time a witness comes to testify at the trial he has typically ‘presented’ at least some of his information on several occasions, for example in informal conversation, in interviews with the police, with one or more lawyers… [t]his is why it is useful to think in terms not merely of witnesses testifying…but of the creating and processing of information.23
*INTLJEVIDENCEPROOF 106 These encounters will not necessarily constitute the one-way flow of information from witness to other parties on matters relating to the relevant event, which is suggested in this passage. Police officers might, variously or cumulatively, have seen the person suspected of committing the offence; spoken to other witnesses who have provided information about the event and offender; formed their own views as to the guilt of the suspect; spoken to other officers who have done either of these things, etc. These activities may well be necessary during the course of an investigation, nevertheless, the potential for cross-pollination of inaccurate information, the drawing of false inferences, and the resultant risk of error is manifest. Nor are the potential risks restricted to encounters between witness and police officers. Conversations between lawyers who might have seen other witness statements, friends and relatives who might have heard rumours about what had happened and who the culprit might be all give rise to danger that subsequent recollection of events will be coloured by information (and misinformation) acquired from other parties. As the process draws out the opportunity for such corruption increases and the task of detecting it becomes more difficult.
The trial is a sub-optimal mechanism for inquiring into the reliability of identification evidence. It has long been recognised that the traditional safeguards of cross-examination and witness demeanour are of little utility in assessing reliability. Furthermore, in light of the malleable nature of memory, the effectiveness of scrutinising identification evidence at trial is dependent on the extent to which past events can be accurately recounted by various witnesses. Establishing that encounters occurred between police and witness, and what transpired during them is not particularly problematic where the police are placed under a duty to record events. However, imposing a wide-ranging duty to record what occurs during every encounter between a witness and police officer is neither practicable nor desirable given the burden it would impose on police investigations.
Conducting procedures under controlled conditions provides the opportunity for various records to be made by police and those representing the suspect's interests and for an incontrovertible record to be made, for example a video recording of events. In this respect they are capable of providing a ‘snapshot’ of considerable probative value of the reliability of a witness's purported identification of a suspect. They also provide a ‘sterile’ environment, one in which the possibility of bias and suggestion is reduced to a considerable degree, in which we can be satisfied that the identification was made in circumstances which largely minimised the risk of error. If formal procedures provide a reliable snapshot of the accuracy of a witness's identification in relation to the issue of error, what occurs thereafter would appear to be of no significant consequence. *INTLJEVIDENCEPROOF 107 Attention can then be focused, in the period leading up to formal procedures being held, on anything that might have a bearing on reliability.
If formal procedures can be conducted within a relatively short period following the witness's observation of the culprit, a number of benefits accrue and possibilities arise. Not only is the witness's memory for events likely to be more accurate, but there are likely to be fewer opportunities for an encounter between the witness and other parties and, as a result, a reduction in the risk of memory contamination or distortion. Moreover, the imposition of a strict duty to conduct procedures at the earliest opportunity might render the imposition of a duty to record events during this period less burdensome. Of course, this duty would have to be a variable one as there are likely to be circumstances beyond the control of the police which result in long delays in conducting procedure, for example those in which the suspect evades detection or refuses to cooperate with attempts to conduct procedures. So to what extent does English procedure address the concerns and implement the measures considered above?
Pre-trial procedure in England and Wales
The pre-trial procedure concerning eyewitness identification in England and Wales is largely contained in Code D of the Codes of Practice, issued by the Secretary of State in fulfilling the duty imposed by s. 66 of the Police and Criminal Evidence Act 1984. The Code consists of three broad sets of provisions:
1. prescribing how procedures for procuring eyewitness identifications under controlled conditions (formal procedures) are to be conducted;
2. setting out when formal procedures are to be held (i.e. the circumstances in which the police are required to hold them); and
3. a set of provisions regulating the way in which identification evidence should be procured (i.e. by ‘street identifications’, discussed below) before any obligation to hold formal procedures arise.
The Code imposes a duty on the police ‘in the interests of fairness to suspects and witnesses’ to hold procedures as soon as practicable.24 The video identification procedure enjoys primacy in the hierarchy of formal procedures25 established by *INTLJEVIDENCEPROOF 108 the Code. This procedure involves the use of moving images of the suspect in a compilation of images including those of at least eight other people who ‘resemble the suspect in age, height, general appearance and position in life’.26
Code D largely nullifies the threat of suggestion arising during formal procedures. It also imposes various recording requirements. In respect of the formal procedures, the officer conducting the procedure must ensure that witnesses are not able to communicate with each other about the case or overhear a witness who has already taken part in the procedure.27 The officer is prohibited from discussing the composition of the video with the witness28 or the outcome of an attempt at identification by another witness.29 Only one witness at a time is permitted to take part in a procedure.30 Before any attempt at an identification, the officer conducting the procedure is required to instruct the witness that the person seen committing the offence may, or may not, appear in the procedure and if the witness cannot make a positive identification he should say so.31 If a witness has previously viewed any photograph, or computerised/artist's impression, he is not to be reminded of doing so,32 and in any case the witness is to be asked after taking part in the procedure whether he has seen any broadcast of published images or descriptions of individuals suspected of committing the offence.33 There is a duty to record a multiplicity of facts including (1) the first description of the culprit provided by the witness,34 (2) the grounds of any objection by the suspect to the proposed procedures and, if necessary, the reasons why it was not possible to overcome the grounds of the objection,35 (3) anything said by the witness about any identifications or the conduct of the procedure.36 Furthermore, there is a requirement that the conduct of the procedure itself be video-recorded.37
The provisions of the Code concerning how procedures are to be conducted exhibit a degree of sophistication that is probably unparalleled in the criminal procedure of any other common law jurisdiction. In 1999, following a study concerning a *INTLJEVIDENCEPROOF 109 number cases in which DNA evidence had been used to exonerate individuals convicted primarily on the basis of eyewitness testimony,38 the Office of Justice Programs of the US Department of Justice convened a technical working group comprising police officers, legal practitioners and researchers from the field of psychology. The group recognised that in the past, ‘procedures ha[d] not integrated the growing body of psychological knowledge regarding eyewitness evidence with the practical demands of day-to-day law enforcement’.39 Reviewing the recommendations contained in the guide, Eyewitness Evidence: A Guide for Law Enforcement,40 that was produced by the working group, one commentator ventured that with the exception of two of the recommendations41 ‘existing regulations on identification in England and Wales either anticipate these proposed rules or provide safeguards which go beyond their very modest proposals’.42
Notwithstanding such plaudits, there is a danger of adopting too narrow a view in evaluating existing safeguards, the adequacy of which can only be properly evaluated in the context of the process as a whole. It has been argued here that diagnosing mistaken identification becomes increasingly difficult as the legal process draws out. The most effective means of tackling this problem lies in the pre-trial procedures conducted under controlled conditions that are set out in Code D. Unfortunately, the potency of these procedures is undermined on a number of counts. The provisions concerning how the procedures are to be conducted, display an admirable degree of sensitivity to the findings of empirical research. However, there is little point in expending great effort in devising sophisticated diagnostic procedures of intricate detail without devoting similar efforts to legislating carefully for their application and enforcing adherence to the scheme. Similarly, the benefits of conducting procedures under controlled conditions might be undermined if insufficient attention is paid to regulating what happens in the process prior to any question of these being conducted. It is to these problems that we now turn.
*INTLJEVIDENCEPROOF 110 Street identifications
The duty to conduct the formal procedures prescribed in Code D is triggered once the police have a ‘known suspect’; that is, that the police have sufficient information to justify the arrest of a person for involvement in the relevant offence.43 The Code provides some guidance on the steps that may be taken where there is no ‘known suspect’. Street identifications typically occur shortly after the event during which the witness has observed the culprit. Where the police response is swift there exists a possibility that the culprit could be found in the locality. A street identification might arise where the witness, having been taken around the area, purports to identify the culprit among those passing by. It might also occur in circumstances in which officers, other than those who are in the company of the witness, happen across someone who resembles the person described by the witness. If the specificity of the description provided by the witness is such that the likelihood of there being more than one person bearing a resemblance in the vicinity is remote, there might be grounds for arrest on the basis of description alone. However, where the description is more general, the officers might reasonably take the view that an arrest could not be justified without the witness purporting to identify him as the offender. In such circumstances an improptu confrontation would be arranged between witness and suspect for this purpose.
The previous version of the Code (1995) contained inadequate guidance on the procedures to be followed in attempting to procure a street identification.44 The latest version (2003) offers some elaboration, requiring the police to make a detailed record of events as soon as practicable, including, (1) details of the location in which any identification took place (2) how it was made (3) the prevailing conditions, and (4) anything said or done by the witness regarding the identification.45 Where there are several witnesses it requires them to be kept separate and, where practicable, taken to view the suspect independently.46 If one of the witnesses identifies the suspect then formal identification procedures have to be adopted in respect of the witnesses who have not had an opportunity to observe the suspect.47
*INTLJEVIDENCEPROOF 111 As noted, arranging for a witness to attempt to identify a person ‘detained’ by the police is an inherently suggestive procedure and there have been calls from the police for greater guidance on the procedure to be followed in order to mitigate suggestiveness and avert the risk of any evidence being excluded.48 In circumstances in which a person has been stopped by the police, all possible steps ought to be taken to avoid giving the witness the impression either that the person stopped is under arrest or is subject to such a degree of control by the police that he must be strongly suspected by them to have committed the relevant offence. Where practicable, officers standing with a detained suspect ought to be in plain clothes and steps should be taken to ensure that uniformed officers and marked police vehicles are not in the immediate vicinity. Furthermore, where a suspect has been detained but it is asserted by the police that in the absence of a street identification by the witness there would be insufficient grounds to make an arrest, there is no requirement (equivalent to those that exist in respect of the formal procedures) to warn the witness that the person that they are about to see may or may not be the person seen a short time before and that if he cannot make a positive identification he should say so. Given the particularly suggestive nature of these encounters there is a case for issuing an even more emphatic warning.49
It must be appreciated that police officers required to carry out street identifications are placed in an invidious position. The exigency of establishing, for investigative purposes, whether those present at the scene were involved in the offence must be tempered by precautions taken to ensure the reliability of any identification. It has been pointed out that this demands a measure of detached reflection.50 In this respect the provision of adequate guidance would appear to be particularly important. The most recent version of the Code enjoins the police to ‘follow the principles applicable to the formal procedures set out in the Code as far as possible’.51 Implicit in this direction is that the guidance that is *INTLJEVIDENCEPROOF 112 provided is less comprehensive than it ought to be. It is unreasonable to expect officers, who in the very early stages of investigation are required to take prompt and decisive action, to ensure that the steps they take are grounded on principles that must distilled from various provisions of a Code of some length and complexity.
The regimen established in Code D concerning the steps that the police ought to take in these circumstances is inadequate and it is submitted that such encounters are attended by a significant risk of an outbreak of the system strain of the misidentification disease. Whether holding formal procedures following a street identification is effective in screening for unreliable street identifications made by witnesses has proved a rather vexed issue for the English courts52 and one that will not be considered further here.